2010年中国大陆100m分辨率人口栅格数据集

本数据集为2010年中国大陆100m分辨率人口栅格数据集，栅格单元单位为人/公顷。本研究融合地理空间大数据（兴趣点）、多源遥感数据（包括夜间灯光、植被指数等）以及路网等数据，基于机器学习方法-随机森林，模拟了中国大陆人口的空间分布，并与2010年的WorldPop人口数据进行了对比，结果显示，由于融合了兴趣点数据，新的100m人口栅格数据产品比WorldPop的精度更高，显著减少了城市中心的人口低估和农村、郊区的人口高估，相关成果“Improved population mapping for China using remotely sensed and points-of-interest data within a random forest model” 2019年在TOP期刊《Science of Total Environment》发表，并成为ESI前1%高被引论文。This dataset is a 100-meter resolution population grid dataset for mainland China in 2010, with the unit of each grid cell being people per hectare. This study integrated geospatial big data (Points of Interest, POIs), multi-source remote sensing data including nighttime light data, vegetation indices and other datasets, as well as road network data, and simulated the spatial distribution of population in mainland China using the Random Forest machine learning method. It was compared with the 2010 WorldPop population dataset, and the results showed that the new 100-meter population grid dataset has higher accuracy than WorldPop due to the integration of POI data, significantly reducing the under-estimation of population in urban centers and the over-estimation in rural and suburban areas. The related research paper titled "Improved population mapping for China using remotely sensed and points-of-interest data within a random forest model" was published in the top-tier journal *Science of the Total Environment* in 2019, and has been classified as an ESI top 1% highly cited paper.

搜集汇总

背景与挑战

背景概述

该数据集是2010年中国大陆100米分辨率的人口分布栅格数据，采用随机森林算法融合兴趣点、夜间灯光等多源地理空间和遥感数据生成，相比WorldPop数据具有更高精度，有效改善了城市和农村地区的人口估计偏差。相关研究已发表于国际顶级期刊，并成为高被引论文。